SEARCH

SEARCH BY CITATION

Keywords:

  • targeting peptide;
  • rhabdomyosarcoma;
  • phage display

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma in children. To improve existing therapies and broaden the spectrum of cytotoxic agents that can be used in RMS treatment, we performed a phage-display-based screening for peptides that bind specifically to RMS cells. Two peptides binding to RMS and to other tumour cell lines, but not to normal skeletal muscle cells and fibroblasts, were isolated from phage-displayed random peptide libraries. One peptide, named RMS-I (CQQSNRGDRKRC) contained the integrin-binding motif RGD and its binding was blocked by an antibody against αvβ3integrin, which is expressed on the RMS cell line RD. The isolation of RMS-I confirmed the validity of our screening procedure. The second peptide, named RMS-II (CMGNKRSAKRPC), shows sequence similarity to a previously identified peptide with tumour lymphatic specificity, LyP-1. However, RMS-II binds in vivo to RMS xenografts better than LyP-1 and homes to the tumour blood and not to lymphatic vessels. Therefore, RMS-II represents a promising peptide for the development of RMS-specific targeting approaches. © 2008 Wiley-Liss, Inc.

Rhabdomyosarcoma (RMS) is the most common soft tissue sarcoma in childhood, representing 5–8% of all malignancies in children.1 The tumour is believed to originate from muscle precursor cells. On the basis of histological criteria, RMS tumours are classified into 2 major subgroups, namely the more frequent embryonal RMS (eRMS, 60%) and the aggressive alveolar RMS (aRMS, 20%).2, 3

Treatment response and prognosis of eRMS and aRMS vary widely depending on location and histology, even though aRMS is generally the more aggressive subtype with the poorest prognosis at diagnosis.4 Therapy consists of a combination of surgery, chemotherapy (vincristine, actinomycin D and ifosfamid) and local radiation in the case of aRMS.4 Whereas improved chemo- and radiotherapeutic regimens have increased the 5-year event-free survival of eRMS patients to approximately 67%, the 5-year event-free survival of patients with alveolar histology is still low at approximately 41%, because of a greater frequency of disseminated metastases.3 These poor long term survival figures clearly indicate that there is a need to develop better therapeutic options for RMS treatment.

Tumour targeted drug delivery has raised a lot of interest as a way to improve current therapies.5 The results have been promising and peptide-coated liposomes have been used to deliver therapeutic molecules such as cytostatic drugs.6 The best studied examples are peptides containing the tripeptide RGD, which target the tumour and its vasculature.7 These have been used to deliver cytostatic drugs directly to the tumour,6 as liposomal formulations,8 conjugated to nanoparticles9 or as radiolabelled tracers.10, 11

Phage display libraries are used to select defined peptide sequences interacting with a particular target molecule.12 Phage libraries containing up to 1010 different peptides can be constructed in a short time. Since the first description of the isolation of a tumour targeting peptide by phage display, many different targeting peptides have been identified (reviewed in Ref.13). However, no screening has been performed for RMS to date.

In this work we applied phage display technology on a RMS cell line to identify peptides binding specifically to RMS tumour cells, and not to normal cells, both in vitro and in vivo.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Cell lines

The human cell lines RD14 (eRMS), Kelly15 and SH-SY5Y16 (both neuroblastoma), HT-108017 (fibrosarcoma), and MRC-518 (fibroblasts) were obtained from ATCC Type Culture Collection (LGC Promochem, Molsheim Cedex, France); Rh419 and Rh3020 (aRMS) were generously provided by P. Houghton (St. Jude Children's Research Hospital, Memphis, TN); and A36521 (melanoma) was obtained from the Department of Dermatology, Zurich University Hospital (Zurich, Switzerland). These cells were maintained under proliferating conditions in high glucose DMEM medium supplemented with 10% foetal calf serum (FCS; Bioconcept, Allschwil Switzerland) in a humidified incubator with 5% CO2 at 37°C. Normal human skeletal muscle cells SkMC-c were obtained from PromoCell (Heidelberg, Germany), and cultured in Ham's F10 nutrition mixture supplemented with 15% FCS and 2.5 ng/ml human basic fibroblast growth factor 2 (Sigma-Aldrich, Buchs, Switzerland) in 10% CO2 at 37°C. All media were from Gibco(Invitrogen, Basel, Switzerland) and contained 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen).

Phage libraries

Phage libraries were constructed in the T7 phage as described.22 Oligonucleotides coding for cyclic random peptides (length between 7 and 10 amino acids) were ordered from Sigma-Genosys (Pampisford, UK). The codon NNK (where N is T, A, G or C nucleotide and K is either G or T) was used to code for a random amino acid. Double stranded DNA oligos with HinDIII and EcoRI restriction sites were cloned in the T7 phage arms of T7Select 415-1b (Novagen-EMD Biosciences, Darmstadt, Germany). A mixture of 4 cyclic libraries CXnC (n = 7, 8, 9 or 10) was used which displayed 415 copies of the peptide on the capsid. Combined average diversity of the libraries was 1 × 108. The non-recombinant T7-phage (T7-Select Control, Novagen) was used as negative control. Individual phages were constructed by cloning of double stranded oligos coding for a known peptide into the vector arms of T7Select 1-1b (1 displayed peptide on the phage capsid), T7Select10-3b (10 displayed peptides) and T7Select 415-1b (415 displayed peptides) as described above.

Phage selection and validation

RD cells (106) were detached from the culture plate with 2.5 mM EDTA in PBS and reconstituted by incubation in DMEM/1%BSA for 20 min at RT. 2 × 109 plaque forming units (pfu) of the phage library were added to the cells, whereby the volume of added phage never exceeded 1/10 of the total volume. Cells were incubated in 1 ml DMEM/1%BSA by rotation for 2 hr at 4°C, washed 5 times by centrifugation, and finally lysed for 20 min on ice by addition of Nonidet-40 (NP-40) ad 1%. To rescue the bound phages, 1 ml of logarithmically growing bacterial culture BLT5403 (Novagen) was added. Rescued phage was titrated, reamplified and filtered through a 0.2 μm non-pyrogenic membrane. Quantification of single phage binding was performed in the same way. For sequencing, phage clones were picked randomly from single plaques and PCR amplified as previously described.22

Peptide and antibody competition

Synthetic peptides were synthesized and purified by Eurogentec (Seraing, Belgium) to 95% purity. Peptides were circular through a cysteine bridge and a biotin was added at the N-terminus. Peptide stock solutions (2 mM) were prepared in PBS and stored at −20°C. Peptide RGD-4C was obtained from AnaSpec (29897; San Jose, CA). The mouse anti-human integrin αvβ3, clone LM609 was obtained from Chemicon International (MAB1976Z; Millipore, Zug, Switzerland). For competition experiments 106 cells were first detached from the culture plate with 2.5 mM EDTA in PBS and reconstituted by incubation in DMEM/1% BSA for 20 min at 37°C. Cells were then incubated 10 min under rotation at room temperature with the peptide or antibody in 100 μl volume. Subsequently, phage (2 × 109 pfu) in 900 μl DMEM and 1% BSA were added to the cells and incubated under rotation for 2 hr at 4°C. After 4 wash steps with DMEM/1% BSA and treatment with 0.1% NP-40 for 20 minutes on ice, phage titers were analysed by infection of BLT5403 and calculated over non-recombinant phage T7.

In vivo experiments

CD1-Nu/nu mice (Charles River, Germany), 4 to 6 weeks old, were injected s.c. into the dorsolateral flank with 5 × 106 RD cells to produce RMS xenografts. Phage (109 pfu in 200 μl) or peptides (200 μg in 200 μl) were injected into the tail vein (i.v.) of mice bearing size matched tumours (750 mm3). After 10 min of circulation, mice were perfused with 50 ml PBS through the left heart ventricle to remove unbound phage or peptide. Tumour and control organs were dissected, weighted and washed with ice-cold PBS. For quantification of phage binding, tissue samples and the tumour were homogenized with a Medimachine (Becton Dickinson, Allschwil, Switzerland) and phages were rescued by bacterial infection after lysis of the cells in 1% NP-40 as described.22 Blood vessels were visualized by i.v. injection of Texas Red-labelled Lycopersicon Esculentum (tomato) lectin (Vector Laboratories, Reactilab SA, Servion, Switzerland) as described.22 For microscopic analysis, tissues were embedded in TissueTek O.C.T. compound (Sakura Finetek Europe B.V., The Netherlands), frozen, sectioned at 6 μm and fixed in cold acetone for 10 min. For visualization of lymphatic vessels, the Syrian hamster anti-mouse podoplanin monoclonal antibody 8.1.1 (Developmental Studies Hybridoma Bank, University of Iowa, Ames, IA) was used at a dilution of 1:500 overnight at 4°C, and detected with an anti-hamster IgG Alexa594 (Invitrogen) antibody at 1:200. For visualization of biotin-labelled peptides, tissue sections were incubated with 1 μg/ml Avidin-Alexa488 (Invitrogen) 60 min at room temperature, washed and mounted with Vectashield mounting media with DAPI (Vector Laboratories).

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Enrichment of RMS-binding phage by in vitro screening

To identify peptides binding specifically to RMS cells, we performed an in vitro phage display screening with 4 cyclic random peptide libraries on the embryonal RMS cell line RD (Fig. 1a). The length of the random peptides varied between 7 and 10 amino acids, which represents a compromise between short sequences with statistically few stop-codons, and long sequences with more potential binding sites but also more stop-codons and truncated linear peptides. The embryonal histological variant of RMS was chosen because it represents the majority of RMS cases.23 Cultured cells were incubated in suspension with 1 × 109 plaque forming units (pfu). After washings, phage bound to RD cells were rescued by bacterial infection. Enrichment of bound phage was monitored by comparing the binding of selected phages to that of non-recombinant T7 phage (T7). After 3 rounds of selection on RD cells, a 52-fold enrichment was observed which could be further increased to 320-fold after the fifth round (Fig. 1a). The last round of selection was repeated and a similar enrichment was observed (not shown), confirming the successful selection of phages binding to RD cells after 5 rounds of selection.

thumbnail image

Figure 1. Selection of phage-displayed peptides binding to RD cells. (a) Phage library displaying random cyclic peptides (CX7–10C) was incubated with RD cells, washed, titrated and reamplified. Binding of the phage library was compared with binding of the nonrecombinant T7 phage after each round of selection. (a) A 52-fold enrichment was observed after the third round of selection and could be further increased by 2 additional rounds of selection to 320-fold. (b) Binding of 2 phages enriched after selection on RD cells, RMS-I and RMS-II, was tested on different tumour cell lines and on normal cells (SkMC-c: myoblasts; MRC-5: fibroblasts). Data is shown as fold binding over nonrecombinant T7 phage.

Download figure to PowerPoint

Selected phages bind to tumour cells but not to normal cells

The sequences of the peptides enriched through binding on RD cells were analysed by sequencing of a limited number (n = 16) of randomly selected phage clones from the last round of selection (Table I). Three sequences were found more than once and were named RMS-I (CQQSNRGDRKRC), RMS-II (CMGNKRS AKRPC) and RMS-IV (CESHRQRRAKC). RMS-I contains the tripeptide RGD, a well known tumour targeting motif,24 whereas RMS-II has sequence homologies to LyP-1, a lymphatic tumour targeting peptide recently derived from an in vivo phage screening by depletion of CD31-positive cells from xenografted breast carcinoma.25 RMS-IV showed no previously described motif, but it contains dibasic residues that were present in 4 other phages, indicating a potential targeting consensus for surface molecules like pro-protein convertases (manuscript in preparation).

Table I. Sequences of Phage Displayed Peptides Enriched for Binding to RD Cells
PhagePeptide sequence1Phage frequency
  • 1

    Single letter code for amino acid sequence.

  • 2

    Boxed are tripeptide motifs that have been previously identified in tumour targeting peptides. Bold letters indicate amino acid homologies in different peptides. Italic letters indicate recurrent strong polar dibasic residues.

RMS-Iequation imageRKRC23/16
RMS-IIequation image SAKRPC3/16
RMS-IIICKRTSKCGGKC1/16
RMS-IVCESHRQRRAKC2/16
RMS-VCLRKRRENTKC1/16
RMS-VICHHWTFRKTTC1/16
RMS-VIICSPNNTRRPNK1/16

Next, we analysed the binding efficiency of RMS-I and RMS-II to RD cells. When tested individually, both phage clones bound to RD cells two-fold better than the phage pool in round 5 of selection (Fig. 1b). RMS-I and RMS-II bound approximately 540- and 700-fold better than the non-recombinant T7, respectively. These results confirm that the selected phages bind efficiently to RD cells.

To verify the tumour specificity of the selected phages, we tested the binding of individual phages on 2 normal cell types, fibroblasts (MRC5) and myoblasts (SkMC-c), and a panel of tumour cell lines, including the aRMS cell lines Rh4 and Rh30, the neuroblastoma cell lines Kelly and SH-SY5Y; the fibrosarcoma cell line HT-1080, and the melanoma cell line A365 (Fig. 1b). RMS-I and RMS-II binding to normal myoblasts (SkMC-c) and fibroblasts (MRC-5) was 5- to 10-fold lower than to tumour cells, indicating tumour specificity of both phages. Binding of RMS-I to RMS cell lines was very strong (between 500- and 700-fold over non-recombinant T7), and no difference was observed between the alveolar and embryonal RMS subtype. The analysis of other paediatric tumour cells showed a significant binding affinity to fibrosarcoma (800-fold over T7) and a lower but still significant binding affinity to neuroblastoma cells (300-fold over T7). Among the cancer cell lines tested, only the melanoma cell line A365 was not efficiently bound by RMS-I, indicating that the target receptor for RMS-I is not expressed or is not accessible on these cells.

RMS-II bound 700- and 520-fold better to RMS cells RD and Rh4 when compared with T7, respectively. The binding affinity to Rh30, 1200-fold over T7, was higher than to all other cell lines. RMS-II bound to all the other tumour cell lines tested 300- to 400-fold more than T7. These results indicate that both phage RMS-I and RMS-II bind to a tumour-specific molecule(s) which is not only present on the eRMS cell line used for the screening.

Cognate synthetic peptides reduce phage RMS-I and RMS-II binding to RD cells

To validate the peptide sequences and to confirm that the phage binding to RD cells is indeed mediated by the displayed peptides, RMS-I and RMS-II cognate peptides were synthesized and used at increasing concentrations to compete the binding of RMS-I and RMS-II phage. Peptides were first incubated with the cells and then equal amounts of phage were allowed to compete for binding. After washing, bound phages were quantified (Figs. 2a and 2b). Peptide RMS-I was able to compete the binding of the phage RMS-I by 50% at a concentration of 10 nM. The efficiency of the competition did not increase at higher concentration. Peptide RMS-II was able to reduce the binding of phage RMS-II by 75% at the lowest concentration tested (10 nM) and increasing peptide concentrations did not increase the extent of competition. These results indicate that the cognate synthetic peptides retain the RD binding potential of the phages to a certain extent, but are not able to compete phage binding completely. RMS-I and RMS-II bind 500- and 700-fold better to RD cells, respectively, than the non-recombinant T7 in the absence of peptide competition. At the maximum concentration there is a residual phage binding which is still 50- to 100-fold higher than binding of the non-recombinant T7. This could be due to the fact that phages RMS-I and RMS-II display 415 copies of the peptide on their surfaces and therefore benefit from the multivalency effect.

thumbnail image

Figure 2. Validation of RMS-I and RMS-II peptide sequences. (a, b) Binding specificity of selected peptides to RD cells was tested by competition with the cognate synthetic peptide. Increasing peptide concentrations were added to fixed amounts of phage (2 × 109 pfu) and 106 RD cells. (a) Competition of phage RMS-I binding with the synthetic cognate peptide RMS-I. (b) Competition of phage RMS-II binding with the synthetic peptide RMS-II. (c) Effect of the number of phage-displayed peptides, or sequence linearization on the binding of phage RMS-I to RD cells. (d) Effect of the number of phage-displayed peptides, sequence linearization, or mutation N4E (asparagine to glutamate) on the binding of phage RMS-II to RD cells, and comparison with LyP-1 phage binding. (e) Competition of RMS-I (10 copies) binding to RD cells with RGD-4C peptide. (f) Competition of RMS-I (10 copies) binding to RD cells with anti-αvβ3-specific antibody (clone LM609). (cf) Results are normalized to the binding of the nonrecombinant T7.

Download figure to PowerPoint

To verify this hypothesis, we generated RMS-I and RMS-II phage displaying 1 copy or 10 copies of recombinant peptides on their surfaces, and compared them with the phage displaying 415 copies (Fig. 2c and 2d). Phage RMS-I displaying 10 or 1 copies bound to RD cells 12- and 6-fold better than the non-recombinant T7, respectively. Compared with the 500-fold better binding over T7 of the phage RMS-I displaying 415 copies, this represents a significant decrease. The binding of phage RMS-II displaying 10 or 1 copies was decreased to 5- and 4-fold over T7, respectively, which is again significantly less than the 700-fold better binding over T7 of the phage RMS-I displaying 415 copies. To gain additional information, we analysed whether circularization of the peptide sequences via the peptide-terminal cysteines influences phage binding to RD cells. Circular peptides were linearised by mutating the terminal cysteine into the neutral residue alanine. Linearization of the displayed peptides decreased the binding of both RMS-I and RMS-II to RD cells, but to a different extent. Binding of linear RMS-I (415 copies) to RD cells was reduced by 40% (Fig. 2c), whereas binding of linear RMS-II (415 copies) was almost abolished (98% decrease; Fig. 2d).

Taken together, these results confirm that the binding of both RMS-I and RMS-II phage to RD cells is mediated by the displayed peptides and indicate that both synthetic peptides retain the same specificity as the corresponding phage, although both peptides show lower binding affinity to the targets on RD cells. Optimal binding to the target might be achieved only by multimeric peptides.

RMS-I binds to integrin αvβ3 on RD cells via its RGD motif

RMS-I contains the tripeptide motif RGD which binds to integrin isoforms overexpressed on tumours (the most prominent being αvβ3 and α5β1).24 To verify whether binding of RMS-I to RD cells is mediated by the RGD motif we first tested the effect of increasing concentrations of RGD-4C, a synthetic peptide which shows an optimal binding to integrins, on binding of RMS-I to RD cells. We used phage RMS-I displaying 10 copies on its surface which binds to RD cells 12-fold better than T7 (40-times less than RMS-I 415). Indeed, RGD-4C peptide could reduce the binding of phage RMS-I (10 copies) to RD cells by 50% at 0.07 μM, and to background levels (2-fold over T7) at 35 μM (Fig. 2e). Because RD cells express αvβ326, and consistent expression of αv integrin is found in primary, recurrent and metastatic RMS,27 we next tested whether phage RMS-I binds αvβ3 on RD cells by incubating cells with increasing concentrations of an anti-αvβ3-specific antibody and monitored binding of RMS-I (10 copies) (Fig. 2f). Increasing anti-αvβ3 antibody concentrations showed a dose dependent inhibitory effect, decreasing binding of phage RMS-I by 3-fold to 3.5-fold over T7 at the highest concentration tested. Taken together, these results show that RMS-I binds to αvβ3 integrin on RD cells via the RGD motif.

RMS-II shows sequence homologies to tumour lymphatic targeting peptides

RMS-II is homologous to Lymphatic Peptide 1 (LyP-1), a peptide derived from an in vivo phage screening on breast tumour xenografts that accumulates in lymphatic tumour vessels but not in the lymphatics of normal tissues25 (Table II). The exchange of asparagine (N) in the NKR sequence of LyP-1 abrogated phage binding to target cells.28 To test whether the RMS-II peptide selected on RD cells might reflect this behaviour of LyP-1, we created a mutant of RMS-II in which the asparagine at position 4 of the peptide sequence was changed into glutamate (E). We then assessed binding of 415 RMS-II phage and N4E 415 RMS-II phage to RD cells in comparison to the binding of a phage bearing 415 copies of the LyP-1 sequence (Fig. 2d). LyP-1 phage did bind RD cells 9-fold less than RMS-II, but 100-fold better than T7. However, mutation of glutamate in N4E RMS-II abrogated the binding to RD cells completely. These results suggest that RMS-II and LyP-1 might share the same binding mechanisms.

Table II. Sequence Homology between RMS-II and Tumour Lymphatics Targeting Peptides
PhagePeptide sequence1Ref.
  • 1

    Single letter code for amino acid sequence.

RMS-IICMGNKRSAKRPC 
LyP-1CGNKRTRGC25
LyP-1bCNKRTRGGC28
LyP-2CNRRTKAGC35

Phage RMS-II homes to tumours in vivo

Next, we focused on RMS-II because of its homology with LyP-1 which shows antitumour activity in other tumour models.28 To test whether RMS-II phage is able to home to tumours in vivo, we generated RMS xenografts by injecting RD cells into the flank of immunocompromised mice. To verify that the structure bound by the phage on cultured cells is retained by the tumours in vivo, we tested the binding of RMS-II phage to single cell suspension derived from RD tumour xenografts. RMS-II phage indeed bound to ex vivo RD cells more than 5,000-fold better than non-recombinant T7 phage (not shown). These results suggest that the target bound by RMS-II is present and even enriched in RD cells grown as xenograft in mice. Next, we tested the in vivo distribution of phage RMS-II and compared it to phage LyP-1 (Fig. 3). After i.v. injection, 10 min of circulation, perfusion and dissection of control organs, the binding of phages to tumour, muscle and brain was quantified. RMS-II phage homed to tumours 30-times better than T7 phage. RMS-II phage binding to RMS xenografts was 10-fold higher than to brain and 3.5-fold higher than to muscle. Binding of LyP-1 phage to tumours was 9-fold over T7 phage, but only 3-fold better than to brain. These results confirm the binding selectivity of RMS-II for RMS-tumours in vivo, and indicate that, for RMS tumours, RMS-II is a better homing peptide than LyP-1.

thumbnail image

Figure 3. In vivo distribution of RMS-II and LyP-1 phage in RD tumour xenografts. Equal amounts of phage were injected into the tail vein of mice bearing RD tumour xenografts. After 10 min of circulation and perfusion, phage binding to tumour, muscle and brain was quantified. Shown are phage output normalized over phage input and tissue weight. Results from 2 independent experiments are shown normalized over control phage.

Download figure to PowerPoint

RMS-II peptide homes to perivascular region in vivo

To confirm the tumour targeting ability of RMS-II and to qualitatively analyze the peptide distribution in vivo using fluorescence microscopy, biotinylated RMS-II peptide was injected into mice bearing RMS (RD) tumours. To specifically visualise the blood vessels, Texas red-labelled tomato lectin (TRL) was injected. After circulation, mice were perfused with PBS to remove unbound peptides, and tumours and control organs were excised. Tumour sections were stained with avidin-Alexa488 (A488) to visualise the biotinylated peptides and examined by fluorescence microscopy (Fig. 4). The RMS-II peptide could be clearly detected in tumour sections (Figs. 4a4c), whereas no fluorescence was visible in control tissues such as brain (Fig. 4d), or muscle (not shown). The peptide signal co-localized with the injected TRL, indicating an association with blood vessels (Fig. 4b and 4c), and did not colocalise with lymphatic vessels positive for podoplanin (Fig. 4a). In control mice, injected with an unrelated biotinylated peptide, no peptide was detected either in the tumour (Fig. 4e4g) or in the brain (Fig. 4h), indicating that RMS-II binding to RMS tumour cells is specific and not due to the biotin modification or unspecific A488 staining. Taken together, these results indicate that, in vivo, RMS-II peptide binds to the tumour associated blood vasculature as well as to tumour cells, and does not home to lymphatic vessels in this RMS xenograft model.

thumbnail image

Figure 4. In vivo distribution of RMS-II peptide. Biotinylated RMS-II peptide (ad) or unrelated biotinylated control peptide (eh) were injected i.v. into mice bearing RD tumour xenografts. After 10 minutes of circulation and perfusion, tissues were collected. Fresh frozen tissue sections were stained with avidin-Alexa488 (A488). Blood vessels (bd, fh) were visualized by co-injection of Texas red-conjugated Lycopersicon Esculentum lectin (TRL). Lymphatic vessels (a, e) were visualised by staining with a podoplanin antibody. (a) Podoplanin staining confirms that lymphatic vessels (red) are present in RMS tumour xenografts and RMS-II peptide (green) does not colocalise with RMS lymphatics. (b) RMS-II peptide (green) colocalises with tumour blood vessels (red) visualised by lectin co-injection. (c) Detail of RMS tumour from a mouse injected with RMS-II peptide, showing a larger blood vessel with peri-vascular localization of the RMS-II peptide. (d) Brain of mice injected with RMS-II (green) and lectin (red) for visualization of blood vessels does not show any accumulation of RMS-II peptide. (eh) Tumour and brain of mice injected with control peptide show no peptide accumulation. Magnifications are as indicated.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

By performing a phage display screening on a RMS cell line we have identified and characterised 2 novel peptides: RMS-I which may bind to αvβ3 on RD cells through the integrin binding motif RGD; and RMS-II which binds to RMS and other cancer types invitro with a greater affinity than to normal cells, and homes to tumour blood vessels as well as RMS tumour cells. Despite a striking sequence homology to the tumour lymphatics targeting peptide LyP-1, RMS-II does not home to tumour lymphatics of RMS.

After 5 rounds of in vitro selection on RD cells, sequencing of phage clones revealed 3 enriched clones. Together, RMS-I, RMS-II and RMS-IV represented 50% of all phage sequenced. RMS-I and RMS-II were selected for further studies and showed a greater specificity for binding to RMS cells when compared with normal cells. RMS-I and RMS-II also showed efficient binding to other tumour types (neuroblastoma and fibrosarcoma), indicating that the targets are not restricted to eRMS cells. Both RMS-I and RMS-II were validated by competing phage binding with cognate synthetic peptides. Here, it has to be noted that competition of phages bearing 415 copies of the peptide on their surface was not complete even at the highest peptide concentration tested. Because binding of RMS-I phage bearing 10 copies of the peptide could be reduced to background levels, we hypothesise that binding of RMS-I, and of RMS-II, benefits from the multivalency effect. They will probably perform best if multimerized or conjugated to the surface of liposomes or nanoparticles.

RMS-I

RMS-I binding to RD cells might be mediated by the RGD motif binding to integrin αvβ3, as competition with the RGD-4C peptide and an antibody specific for αvβ3, expressed on RD cells26 was possible. The selection of RMS-I containing this RGD-motif on the one hand confirms the validity of our screening procedure and on the other reiterates that RGD-integrins-based tumour targeting approaches should be considered also for RMS patients. The RGD-motif was first identified as the minimal motif mediating adhesion of fibronectin to integrins.29, 30 Subsequently, an RGD containing phage has been isolated from an in vivo phage display screening on MDA-MB-435 breast carcinoma xenografts.7 Importantly, RGD-motif containing peptides have been shown to be able to deliver drugs to tumours in vivo7. RMS tumours express a different pattern of integrins31 than breast tumours and αν is consistently expressed in primary, recurrent and metastatic RMS.27 Insertion of an RGD sequence into the fibre knob of Coxsackievirus-adenovirus increases infection of RD cells, mediated by αν integrin.26 Despite the wealth of information about integrin involvement in cancer in general, and in RMS32, 33 specifically, nothing is known about the potential of RGD-mediated integrin targeting for RMS treatment. Therefore, our results offer a rationale to test RGD-based tumour targeting agents like the therapeutic Cilengitide34 for the treatment of RMS.

RMS-II

RMS-II has a striking sequence homology with several previously described tumour lymphatics targeting peptides (Table II). LyP-1 was selected from a cell suspension prepared from MDA-MB-435 breast carcinoma xenograft after depletion of CD31 positive cells25 and LyP-2 was identified from a screening on a transgenic model for squamous cell carcinoma.35 LyP-1b is closely related to LyP-1, and strongly accumulated to MDA-MB-435 tumours.28 The sequence homologies, and the fact that mutation of the glutamate abrogated binding of RMS-II and LyP-1 to tumour cells, could indicate the existence of a tumour lymphatic targeting motif consisting of the NKR sequence, shared by RMS-II, LyP-1 and LyP-1b. LyP-2 has a weaker sequence homology to RMS-II, LyP-1 and Ly-1b (NxRTxxG). LyP-2 binds to tumour lymphatics of skin and cervical cancers of the K14-HPV16 transgenic mice but not of MDA-MB-425 breast tumour,36 which are efficiently bound by LyP-1. When tested in vivo, RMS-II peptide did not home to tumour lymphatics of the RD RMS xenografts, and RMS-II phage binding to RD RMS xenografts was two-fold higher than LyP-1. Therefore, NKR does not seem to represent a lymphatic targeting motif, indicating a tumour type selectivity for these targeting peptides.

RMS-II phage homed to tumours 10-fold better than to brain, and 3.5-fold better than to muscle. RMS-II peptide accumulated in tumours but not in other organs studied (brain, muscle, skin). Interestingly, histological analysis of the tumours suggest that RMS-II accumulates on blood vessels and in the peri-endothelial region of tumours, implying that tumour blood vessels might have increased expression of the target receptor for RMS-II, and tumour blood vessel leakiness allows direct access to tumour cells and accumulation of RMS-II. The involvement of lymphatic vasculature in RMS has not been studied systematically. We have detected significant expression of VEGF-A and -C, Ang2 and PROX-1 by quantitative real-time PCR in RMS cell lines and RMS tumour biopsies (K. Iljin, M. Bernasconi, unpublished results). All of these genes have been shown to participate in lymphatic vessel development. Whereas VEGF-A is also a major regulator of blood vasculature, VEGF-C and PROX-1 are more selective regulators of lymphatic vasculature. Histological analysis of RMS xenografts confirmed these observations, showing podoplanin-positive lymphatic vessels. However, RMS-II peptide was never found in podoplanin positive vessels. We therefore can exclude that the different behaviour of RMS-II and LyP-1 is due to a lack of tumour lymphatics in our model. Rather, these results reinforce the notion that lymphatic vessels, just like vascular blood vessels, exhibit different molecular characteristic depending on the tumour type. Despite their sequence homologies, RMS-II, LyP-1, LyP-1b and LyP-2 seem to bind distinct targets in tumours. The receptor for LyP-1 has recently been reported.37 LyP-1 binds to p32/gC1qR, a mitochondrial protein transported preferentially to the cell surface under hypoxic conditions. Despite our efforts we have not been able to identify the receptor for RMS-II. LyP-1 peptide has been shown to possess antitumour activity in vivo28, and therefore represents a particularly interesting tumour targeting tool. RMS-II peptide does not reduce RD cell viability and therefore must be conjugated with cytotoxic drugs or particles with anti-neoplastic properties to have a therapeutic effect (H. Witt, M. Bernasconi, unpublished observation).

In conclusion, RMS-II peptide represents a promising candidate for the development of RMS-specific targeting. Only the comparison with RGD-based integrin targeting tools, e.g., RGD-4C-based molecular tracers, will show whether RMS-II really offers an advantage for RMS and for other types of tumours.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References